Stress Linearization

calebvoisin

I am testing out Mecway for the first time. I have tried a simple pipe which I created in Inventor and saved as an IGS file then imported to Mecway. I added a fixed support on one end, internal pressure, and a tension load on the opposite end of the fixed support. After running, I tried to perform stress linearization, but when I do it tells me that some nodes in the selection don't have 3D stress state. Not sure what I'm doing wrong.
Thanks for the help.

Sergio

Could it be that as the model is IGS you have only a shell mesh? Maybe if you import as STEP Mecway will produce a solidmesh. Or use the Mesh Tools\Automesh 3D on your existent shell mesh. Guess that you must check node conectivity first in the second case.

Regards

calebvoisin

thank you. Once I used a STP, it worked.
CV

Andrea

Which theory did you follow for calculating SCL Victor?

Victor

Not sure if it has a name but these are the formulas it uses:

σ = total stress tensor transformed so the x-axis is aligned with the SCL.

t = thickness.

                    t/2

                1   / 

σ_membrane =    -   |  σ dx

                t   /

                   -t/2

Bending stress at the first point

                      t/2

                -6    / 

σ_bending =     ---   |  σ x dx

                t^2   /

                      -t/2

If the "Ignore through-thickness bending stress" option is checked, this integral uses 0 for the 3 components in the direction of the SCL (σ_xx, σ_xy, σ_zx).

σ_m+b = σ_membrane + σ_bending

σ_peak = σ - σ_m+b

Andrea

Ok! It's a classical one (ASME code).
Reading about SCL i found that it can understimate or ovestimate stresses if the stress distribution isn't axisymmetric. The most recent proposal is to consider, for general cases, Stress Classification Plane instead SCL. In this case the user must define the plane and an axis for calculating bending stress.
SCL is typical for pressure vessel design but could be usefull also for general case of studies. I think that SCP will be more usefull.

Example of section of a clamped beam subjected to a pure bending: Upper SCL shows membrane stress equals to maximum bending stress - Transverse SCL shows typical bending stress distribution (absolute values) Max theoretical sigma = 6 MPa

Victor

I see an abstract here http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid=1628315

It sounds like it would be more convenient - don't make so many lines, just a small number of planes? I'm surprised someone's come up with a new method that still meets the code requirements, so it could be applied straight away. I probably won't implement that though unless there's more interest. Even SCL seems to be rarely used.

From what I understand, and I'm probably wrong here, but the whole concept of stress classification is a poor-man's alternative to material nonlinearities, isn't it? Doesn't ASME have the option of doing it that way instead? If that's true, then maybe it's not very useful these days where nonlinear FEA is quite ubiquitous.

Andrea

Maybe you're right. EN 13455 (European standard for pressure vessels) contains a route based on material nonlinearity and this was introduced about 10 years ago but requires good knowledge in FEA (above all to perform Progressive Plastic deformation design check). I think that will be developed during next years

I think that stress categorization is quite important because helps to understand which components of the stress are more important under a different point point of view.
In absence of secondary stresses (membrane+bending) primary membrane and bending stresses are quite near to be "nominal stresses". The problem is that SCL is reliable only for axisymmetric problems.
Natural way to use SCL is through thickness using axisymmetric elements (plane elements)

Another problems is to subdivide primary stresses from secondary stresses.

The possibility of subdividing nominal from peack stress could be necessary for example in fatigue analysis based on nominal stress (faster method).

Obviously non-linear analysis is more near to the real behaviour of the component but stress categorization could be parellel instrument for the designer.

Sergio

Back to this thread, is possible to compute stress linearization on Mecway for shell or axisymmetric elements? Have made some tests last week with no success.

Regards

Victor

It still only works with solid elements. Though that includes the solids generated by CCX from shells and axisymmetric models as long as their nodes are properly connected.